INDEX

Explanations

the presence of the word "help"

oai_token-act-pair · gpt-3.5-turbo

the phrase "can't help" in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

theless

-0.66

etheus

-0.64

andom

-0.59

avior

-0.58

sonian

-0.58

naire

-0.56

punk

-0.54

esome

-0.54

initely

-0.54

ortality

-0.54

POSITIVE LOGITS

ctor

0.66

 noticing

0.64

ãĤ®

0.61

":["

0.60

 Sega

0.59

des

0.58

ãĤ¨ãĥ«

0.58

 fielding

0.56

#$#$

0.55

 sponsoring

0.55

Activations Density 0.033%

the presence of the word "help"

the phrase "can't help" in various contexts

No Comments

No Known Activations